

LLM 101: How ChatGPT Actually Works
You type a question into ChatGPT, and a few seconds later an answer starts appearing. What actually happens in between? And why can the same system explain something brilliantly one moment, then make a glaring error with complete confidence the next?
In this free, live session, we’ll take a single prompt and follow its journey through a large language model to the generated answer. You’ll leave with a clear mental model of how LLMs work, built through simple explanations and live demonstrations, without heavy maths.
Who it’s for
CS students and Software Engineers who use tools like ChatGPT, Claude, Gemini or Copilot and want to understand the systems behind them.
No prior machine learning background is needed. Other curious learners are welcome too.
What we’ll cover
Tokens: How text is broken into pieces, and why those pieces matter for usage costs and limits.
Generating answers: How a model produces its response one token at a time, and what “temperature” changes.
Attention and context: How a model uses information from the text it receives, and what a context window means.
Hallucinations: Why models can produce convincing but incorrect answers, and what that means when using them or building with them.
ChatGPT will be our running example, but the core ideas also apply to Claude, Gemini and other modern LLMs.
How the session runs
Total duration: approximately 2.5 hours, including a break and Q&A.
Part 1 · 60 minutes: From your prompt to tokens and generated text.
Break · 10 minutes
Part 2 · 60 minutes: Intuition behind attention, context, and hallucinations.
Live Q&A · 20 minutes
There will be live demonstrations and opportunities to think through examples together. You can submit a question when registering to help shape the session.
Joining details
The session will take place on Zoom. You'll get a calendar invite as soon as you register. The Zoom link will appear on this page and in your reminder emails before the session.
Recording and Live Stream
The session will be streamed live on YouTube (@engineeringclearly), and I'll publish an edited recording afterwards.
About your host
Tarun Malhotra is a senior software engineer with experience building large-scale systems. He has taught engineering concepts to thousands of learners through various platforms.
Through Engineering Clearly, he shares approachable explanations of how modern software and AI systems work.
This is an independent session hosted by Tarun.
Questions before the session? Email tarun.iiitd@gmail.com.